A short story from an outage that “wasn’t an outage” A fintech API team had green dashboards across the board: uptime healthy, CPU normal, pods running, database latency stable. Yet support tickets were piling up. Transfers were getting “accepted” but…
Author: Ankur Sharma
-
The Release You Couldn’t Prove: GitHub Artifact Attestations, npm Provenance, and a Deploy-Time Verification Runbook
GitHub artifact attestations and npm provenance, explained with a deploy-time verification runbook so you can ship releases with auditable build trust.
-
The Model Gateway Meltdown: An AI/ML Production Blueprint for Capability Drift, Cost Spikes, and Safe Fallbacks
A Saturday incident that looked like “random model weirdness” A team shipped a customer-support copilot on Friday, then woke up Saturday to a mess: summaries got longer and less useful, latency doubled for one region, and token spend jumped 38%…
-
The Late Event That Rewrote Friday: A Data Science Playbook for Watermarks, Incremental dbt Models, and Safe MERGE Backfills
Late-arriving events reconciliation made practical with event-time watermarks, dbt incremental models, and safe MERGE backfills for trustworthy dashboards.
-
The Update That Broke Only Logged-In Users: A WordPress Engineering Guide to Safe Plugin Interoperability in 2026
A bug that hid from everyone except your best customers A membership site pushed a routine release at 11:40 PM. Home page looked fine, checkout worked for guest users, and uptime monitors stayed green. By morning, premium members were angry…
-
The OOM Kill That Wasn’t Random: Linux Memory Pressure Monitoring with PSI, cgroup v2, and Kubernetes MemoryQoS
Linux memory pressure monitoring with PSI, cgroup v2 memory.high, and Kubernetes MemoryQoS to reduce surprise OOM kills without overprovisioning nodes.
-
The Metric Drift You Don’t See Coming: SQL and Data Engineering Patterns for Trustworthy Analytics in 2026
A quick story from a board meeting prep On a Wednesday afternoon, a data team was preparing revenue numbers for a leadership review. The dashboard showed steady week-over-week growth. Finance exported numbers from the billing system and got a lower…
-
The Hero Image Bottleneck: A WordPress Runbook for AVIF, srcset Hygiene, and Safe Thumbnail Regeneration
Practical WordPress image optimization runbook for AVIF, srcset tuning, and safe thumbnail regeneration to improve LCP without breaking quality or caches.
-
The Peripheral You Forgot to Threat-Model: Hardening Node.js Systems Across Cloud, Edge, and Home-Server Reality
A quick story that changed one team’s architecture roadmap A startup running a Node.js media workflow platform had excellent cloud hygiene on paper. Their API services were containerized, secrets were in a managed vault, and CI pipelines required approvals for…
-
The Queue That Never Drained: A .NET Worker Service Playbook for Backpressure, Graceful Shutdown, and Retry Boundaries
Practical .NET worker service graceful shutdown guide with Channels backpressure, Polly retry boundaries, and OpenTelemetry, so queue-driven jobs stop cleanly.
-
When the AI Helper Went Sideways: A Python Engineering Playbook for Deterministic Systems in 2026
A Friday deploy, a Monday rollback, and one painful lesson A SaaS team I worked with added an AI-assisted support feature to their Python backend. The idea was straightforward: summarize tickets, suggest responses, and route priority automatically. The first week…
-
The Hover That Beat the Spinner: A 2026 Playbook for Speculation Rules API, 103 Early Hints, and Faster LCP
Use Speculation Rules API, 103 Early Hints, and fetchpriority together to improve LCP and make web navigation feel instant without rewriting your frontend.